GPU Screen Suddenly Turns One Solid Colour and Freezes Mid-Game? Solid-Colour Freeze, Driver Reset & Render-Layer Diagnosis Guide

Key Takeaways
A full-screen solid colour that freezes mid-game is different from a black screen: the display chain is still delivering content, so the fault sits in the rendering or frame-buffer layer. This guide works through the colour, audio, cursor, and on-screen display evidence, then a nine-step decision tree of driver resets, program swaps, and Event Viewer records to narrow the cause.
The Symptom, Stated Precisely
You are in the middle of a game. The picture suddenly becomes one solid colour filling the whole screen — grey, green, pink, orange, blue, or any other single colour — and it freezes. Nothing moves. The picture does not recover by itself, and no message or notification appears. The game was running normally before this, and the freeze happened part-way through play.
That exact shape matters. It is a real symptom shape reported by users in public discussion forums: under graphics load, the screen freezes into a single colour, sometimes with the audio repeating like a broken record, sometimes with the audio continuing normally. The same card on different days can show a solid-colour freeze on one occasion and a black screen on another. The two shapes can alternate, but each incident has to be diagnosed by the shape it shows on that occasion; a black screen last time does not explain a solid colour this time.
A Solid Colour Is Not a Black Screen
The distinction between a frozen solid colour and a black screen is the foundation of this whole guide, and it has to be stated at the start. A black screen means the monitor is receiving no content, or content it cannot display; the companion article on black screens mid-game starts from the monitor's power light and works up the display chain, because a black screen proves nothing about the content layer.
A solid-colour freeze is the opposite situation. The monitor is receiving content, but the content is a single colour that does not move. That proves the display chain — output encoding, cable, monitor, and panel — is broadly still delivering, and the suspicion moves towards the rendering or frame-buffer layer, where wrong content was produced. The two symptoms therefore need different decision trees, and the tests that decide a black screen are not the tests that decide a solid colour.
This symptom is also different from several look-alikes. Structured artefacts — coloured blocks, a checkerboard pattern, horizontal or vertical lines, garbled text, or corrupted textures that move with the picture content — are a different pattern shape, covered in the article on telling VRAM faults from graphics-core faults. A game that crashes back to the desktop is a different ending, covered in the article on crash-to-desktop. A green screen while playing video is a decoding-colour problem on the video path, covered in the article on video playback problems, and is not this symptom. And a driver timeout that ends with a brief black flash, a recovery, and a notification is the TDR outcome, covered in the TDR article; this guide starts from the premise that the picture does not recover.
The Colour Itself Cannot Convict Anything
This is the most important rule in the article: no single colour corresponds to any particular part. Grey, green, pink, orange, or blue — none of them points to a specific component. In the public forum cases, the same kind of fault has produced different colours, and different causes have produced the same colour. No colour-to-component table is given here, and none should be constructed from this article.
The colour has only one legitimate diagnostic use, and it is a matter of record-keeping: is it the same colour every time, or a different colour each time? The same colour every time suggests a more fixed state; a different colour each time suggests a more random state. Even that is only a clue for the record, not a conviction. Note the colour and the time of each incident, and do not draw a conclusion from a single colour on a single occasion. In particular, do not write a frequency ranking of which colour is most common; no source supports one.
Three Different Audio Signals
As with the companion black-screen article, the audio state is evidence — but here it comes in three different forms, and they must be kept separate. They are not three ways of saying the same thing.
First, the audio may continue normally: the game music or dialogue keeps playing. That means the system and the main game program are still running, and the deadlock is limited to the rendering or display layer.
Second, the audio may repeat in a loop: the same second of audio plays over and over, like a broken record. That means the program or audio buffer is also stuck, and the deadlock covers a wider area, possibly at the application level.
Third, the audio may stop completely. That means the system may be frozen as a whole, or the audio path has died too, and the suspicion moves towards a whole-machine direction. Record which of the three you hear before drawing any conclusion.
The Cursor Test
The mouse cursor is another layering test, and it costs nothing. The hardware cursor is drawn by a separate cursor plane that does not pass through the main frame buffer. When the picture freezes, move the mouse: if the cursor still moves, the main frame buffer is deadlocked but the compositing and cursor layers are still running. If the cursor does not move either, the deadlock covers a wider area. A cursor that still moves must not be written as "the graphics card is fine" — it only narrows the range.
What TDR Is, and the Waiting Window
Windows has a built-in mechanism called Timeout Detection and Recovery (TDR). If a graphics engine does not complete its work within a default limit of 2 seconds, Windows resets the display driver. When the reset succeeds, the picture flickers briefly and then recovers — that is the ending of the TDR article. When the reset fails, or too many resets repeat within one minute, the result can be a blue screen with the code VIDEO_TDR_FAILURE (0x116).
In a solid-colour freeze, the first step is therefore to wait 30 to 60 seconds and let TDR finish its reset. Waiting is not hoping the problem fixes itself; it is the necessary window for the reset to complete. Forcing the machine off immediately loses unsaved progress and interferes with the event record. If the picture recovers by itself during that window, follow the TDR article's path. If it does not recover, continue with this guide's decision tree.
The Nine-Step Decision Tree
Work through these steps in this order. Each step is a controlled test with a readable result. Do not skip ahead to reinstalling drivers, and do not change two things at once.
Step 1: Wait 30 to 60 seconds and press nothing
Give TDR its window to complete a reset. If the picture recovers by itself, follow the TDR path. If it does not recover, continue. Do not force the machine off immediately.
Step 2: Record four things — colour, audio, cursor, and program
Note (a) what colour the screen is, and whether it is the same colour every time; (b) whether the audio is continuing normally, repeating in a loop, or completely silent; (c) whether the mouse cursor still moves when you move the mouse; and (d) whether the freeze happens in only one game or program, or across the whole desktop. Record all four before running any test, because every later test is read against them. If only one program is affected, the program-layer direction is the right one.
Step 3: Ctrl + Alt + Del
This is a system-survival test, not a repair. Microsoft's support material on blank screens in Windows lists this step: pressing Ctrl + Alt + Del may bring up the security screen, from which Task Manager can be opened. In this symptom, the reading is: if the security screen appears, Windows itself is still responding, and the deadlock is limited to the game or rendering layer — open Task Manager and check whether the game program is still running or has stopped responding. If the security screen does not appear at all, the system layer is not responding either, so continue.
Step 4: Win + Ctrl + Shift + B
This is the display-driver reset shortcut documented in Microsoft's support material. Press the four keys together once and wait a few seconds, listening for a beep and watching for a flicker. A successful reset produces a brief flicker and a beep. It does not close programs and does not change the driver version.
If the picture returns, the fault points towards the driver or compositor state — but a reset is not a repair. It clears the immediate symptom without changing the driver version or clearing leftovers, so the driver layer still needs follow-up (see Step 7). If there is no response, continue.
Step 5: The OSD test
Press the monitor's menu button. The on-screen display is drawn by the monitor itself, without passing through the graphics card or the cable. If the menu appears clearly, layered over the solid-colour picture, the panel is alive and can still draw normal content — the problem is upstream, in the content being fed to it. If the menu itself is unclear or cannot be seen, the suspicion moves towards the panel, in the direction of the articles on monitor evidence and panel testing.
Step 6: Swap the program and lower the load — only after a restart, never during the freeze
After restarting, try the same game and the same scene again. Then try a different game or program. Then lower the texture or effects settings by one level and try again. Change only one variable at a time. If only one game is affected, the direction is the program layer. If every game is affected, continue. If lowering the settings stops the freeze from happening, the direction is load-related — memory load, power delivery, or cooling comparisons — and the dedicated articles on those subjects are the right next step, not a deeper dive inside this one.
Step 7: The driver layer — only when the fault repeats
Check the driver version date and whether a driver or Windows update happened recently. Return any overclock or undervolt to the factory default for comparison, and do not add voltage or overclock further to "fix" a freeze; no overclocking instructions are given in this guide, and no generic clock or voltage figures are used as a standard. A clean driver reinstall using Display Driver Uninstaller (DDU) is a last resort, and the DDU process itself is covered in the dedicated articles on this site, not repeated here. Do not call for a driver reinstall before Steps 1 to 6 have been done.
Step 8: Collect the evidence after a manual restart
Open Event Viewer (Windows Logs, then System) and look at the records around the time of the incident: Event 4101 from the source "Display" means the driver timed out and was reset; LiveKernelEvent 141 means a single engine timed out and was reset; a 0x116 blue screen means a reset failed; the Reliability Monitor's LiveKernelEvent records can be matched against the time. Records are clues, not convictions. A single incident, with a long gap before the next, may not need major action. Several incidents in one day, or a rising frequency, is worth following up seriously. Kernel-Power Event 41 means only that Windows detected an unexpected shutdown — quite possibly the forced shutdown you performed yourself — and cannot be read as proof that the power supply has failed.
Step 9: Four possible endings
Each ending has its own next step, and none of them is simply "send it for repair" as the only conclusion.
| Ending | What the test showed | Next step |
|---|---|---|
| Driver state | The reset shortcut or a TDR recovery brought the picture back | Follow up the driver layer: version history, recent updates, overclock returned to default, clean reinstall as a last resort |
| Program layer | Only one program is affected | Check that program's files, settings, and recent updates; do not apply this guide's hardware tests to it |
| Load-related | Lowering settings stops the freeze | Compare memory load, power delivery, and cooling against the machine's own usual readings |
| Not narrowed down | No test changed anything | Stop and seek a professional inspection with cross-testing; do not open the card, change voltages, or flash firmware |
VRAM: A Suspect, Not a Verdict
Discussion in public forums points at VRAM damage as one suggestion for random solid-colour freezes under graphics load, but it is only ever stated as a suggestion, and always with the condition that it be backed by cross-testing: changing drivers, changing slots, and testing against another card. The site's article on VRAM faults states the same principle: a pattern is only a tendency, never a conviction.
That is how VRAM must be written about here. A VRAM direction can only be stated when four things line up together: the freeze happens in different games, lowering the texture or resolution settings changes the behaviour, a clean driver installation does not help, and cross-testing supports it. The words "solid colour" on their own prove nothing about VRAM.
The Screenshot Test Cannot Be Done Here
Readers of the site's article on artefacts visible to the eye but not in screenshots may wonder whether a screenshot can settle which layer is at fault. In a solid-colour freeze, it cannot. While the picture is frozen, the screenshot shortcut may not respond at all, and even if a screenshot is captured, it captures only the frame buffer at the frozen moment — it is not independent evidence of which layer failed. Do not ask the reader to take a screenshot during the freeze as evidence. The OSD test in Step 5 is the fast alternative that does not depend on the frozen system.
Boundary Cases
- The solid colour appears in only one game, while other games are normal: the suspicion moves towards the program layer, and the articles on game crashes and game-file checks are the right direction.
- The picture shows artefacts or flickering for a while before turning solid: this may be the lead-up to the same fault. Record the order — artefacts first, then solid colour, versus solid colour with no warning — because that order is itself a clue when the machine is inspected.
- The sound also stops and the keyboard light stops responding: this is not this article's symptom. The machine may be frozen as a whole, and the full-freeze and power directions are the right ones.
- The machine restarts itself or shows a blue screen after the solid colour: that is the TDR reset-failure direction, the 0x116 ending of the TDR article, not this one.
- A green screen appears only while watching video or in a browser: that is the video-decoding path, covered in the article on video playback problems; a green video screen is not this symptom.
- The game shows the message "rendering device has been lost" and then closes: that is a driver, overclock, or API direction, not a full-screen solid-colour freeze. The game closes by itself and returns to the desktop, which is a different ending.
- The solid colour or artefacts are already visible on the startup or BIOS screen: that is the direction for faults present before the operating system loads, not mid-game.
- The solid colour appears on only one monitor or only one connector: the output-chain direction, where swapping connectors is the deciding test. Do not conclude the graphics card is at fault from a single monitor.
- The monitor's own menu is unclear or invisible: the panel direction, in the articles on monitor evidence and panel testing.
- On a laptop: the built-in screen and the external screen must be discussed separately. Do not teach opening a laptop.
- The colour is different every time: the state is more random. Record each colour and time, and do not draw a conclusion from any single colour.
Common Mistakes
- "Grey means the VRAM is faulty, green means the driver is faulty." Wrong. Colour does not correspond to a part. The same kind of fault can produce different colours, and different causes can produce the same colour. Colour can only be recorded, never used to convict.
- "A solid-colour freeze means the monitor is broken." Wrong. A clear OSD means the panel can still draw normal content; the solid-colour content comes from the rendering or frame-buffer layer, and the monitor is only displaying what it is given.
- "A solid-colour freeze means the graphics card has burned out." Wrong. Driver state, program instructions, overclocking, memory load, cooling, and power delivery can all deadlock the rendering layer. No such conclusion can be drawn before the Step 6 and 7 tests.
- "A screenshot can prove which layer the solid-colour freeze is in." Wrong. During the freeze the screenshot test cannot be performed reliably, and a captured frame is only the frozen frame buffer, not independent evidence. Do not ask the reader to take a screenshot during the freeze.
- "Win + Ctrl + Shift + B is a repair." Wrong. It only resets the display driver state. It does not change the driver version or clear leftovers. If it has to be used repeatedly, the underlying cause has not been addressed.
- "Event 4101 appearing means the graphics hardware is faulty." Wrong. Event 4101 only records that the driver timed out and was reset. The cause could be the driver, power delivery, heat, overclocking, or a program instruction; frequency and controlled tests are needed before any direction can be stated.
- "Raising the TDR delay in the registry fixes freezes." Wrong. Microsoft describes the TDR registry settings as for testing and debugging, not as a routine fix. Lengthening the limit only delays the trigger; it does not address the underlying cause. This guide does not teach or recommend registry changes.
- "Reinstalling the driver will fix it." Wrong. Reinstalling the driver only helps if the fault is in the driver layer. Reinstalling before the waiting, audio, cursor, OSD, and program-swap tests have been done is blind trial, not diagnosis.
- "Forcing the machine off does not matter, just hold the power button." Wrong. Forcing the machine off is a last resort. It loses unsaved progress and interferes with the event record; waiting, the reset shortcut, and Ctrl + Alt + Del should be tried first.
What This Article Cannot Conclude
The exact source of the solid colour in any given freeze is not settled by public material: whether the frame buffer was cleared and filled with a colour, the memory contents were corrupted, a display-mode switch failed, or the picture is an intermediate state during a driver reset. This guide states what is known — the symptom shape and the test logic — and does not pick one of those as a confirmed mechanism.
Without opening the machine and running cross-tests, no single solid-colour freeze can be attributed to a particular part. This guide can only say which direction the evidence points, never which part has failed. No driver version or game can be singled out as especially likely to cause this symptom unless the manufacturer has published a known-issue document; the only legitimate check is whether an update happened recently. And the same colour every time cannot be written as proof that a fixed part has failed — colour consistency is only a record-keeping clue, not a conviction.
Frequently Asked Questions
My screen turned one solid colour and froze mid-game. Is my graphics card broken?
Not necessarily. A solid-colour freeze means the rendering or frame-buffer layer produced wrong content, but the cause could be the driver, the program, an unstable overclock, memory load, cooling, or power delivery. Work through the decision tree — wait, record the colour, audio, and cursor, try Ctrl + Alt + Del and the driver reset shortcut, then the program and driver tests — before drawing any conclusion.
Does the colour tell me which part has failed?
No. No colour corresponds to any particular part. The same kind of fault can produce different colours, and different causes can produce the same colour. The only use of the colour is in the record: whether it is the same colour every time or a different colour each time.
The audio keeps repeating the same second over and over. What does that mean?
A looping audio repeat means the program or audio buffer is also stuck, so the deadlock covers a wider area than the rendering layer alone, possibly at the application level. That is different from audio continuing normally, which means the system and game are still running and the deadlock is limited to rendering or display.
Can I take a screenshot to prove which layer is at fault?
No. During the freeze the screenshot shortcut may not respond, and even a captured screenshot only shows the frame buffer at the frozen moment, which is not independent evidence. Use the monitor's on-screen display test instead: if the menu appears clearly over the solid colour, the panel can still draw normal content.
What does Event 4101 in Event Viewer mean after a solid-colour freeze?
Event 4101 only records that the display driver timed out and was reset. It is a clue, not a conviction. Its cause could be the driver, power delivery, heat, overclocking, or a program instruction. Frequency matters: a single isolated record is different from several in one day.
When should I seek a professional inspection?
When every test in the decision tree changes nothing, when the freeze happens several times a day and the frequency is rising, or when the system itself has stopped responding. Bring the notes you recorded: the colour each time, the audio state, whether the cursor moved, whether Ctrl + Alt + Del or the reset shortcut changed anything, and the Event Viewer records around the time of the incident. Our shop is at Room 1209, Dali Plaza, Shek Mun; telephone 2186 8010; WhatsApp 6558 6806. There is no home-visit service. Diagnosis is charged within the published three-tier pricing of HK$400 to HK$2,800.